The Agent Skills Directory

[PROMPT_INJECTION]: The skill metadata contains misleading information by claiming the author is 'github', which is an incorrect attribution for a skill authored by 'fellipeutaka'.
[EXTERNAL_DOWNLOADS]: The skill installs the gh-stack extension from a GitHub repository, which is a well-known service.
[COMMAND_EXECUTION]: The skill instructs the agent to execute various gh and git commands for repository configuration and pull request management.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface by interpolating untrusted data into shell commands.
Ingestion points: The skill processes branch names and commit messages provided by the user during initialization and modification of stacks (SKILL.md).
Boundary markers: No explicit delimiters are used to wrap or sanitize external content.
Capability inventory: The skill executes powerful shell commands (e.g., gh stack, git commit) that depend on the ingested data (SKILL.md).
Sanitization: The instructions state that branch names are used verbatim, with no mention of input validation or escaping.

gh-stack